In an ATM (asynchronous transfer mode) network in which a standby route is set in addition to an active route to be normally used when route calculations are made based on topology information exchanged between the nodes, the present invention particularly relates to an active/standby routing system capable of reducing time required for setting a standby route again for reasons such as a trouble in a bypass in order to allow the active route to be recovered from a trouble.
In a conventional active/standby routing system in an ATM network of this type, there is provided a centralized management apparatus for managing the network topology on a centralized basis. Such a centralized management apparatus collects network topology information from all nodes to calculate a standby route in advance in order to allow a connection to be recovered from a trouble with a small delay.
When the centralized management apparatus sets an active connection, it also sets a standby route for the active connection. A system has been employed wherein, in the case of a trouble, a switching control signal is transmitted from a centralized controller of the centralized management apparatus to each node to switch the connection at each node.
Further, a centralized management apparatus may employ a system wherein no standby route is set in advance and, in the case of a trouble, a route to avoid the trouble is calculated to set a connection to bypass the same.
In a network having a configuration as described above, since the centralized management apparatus can manage information on the topology of the network, a standby route for an active route can be determined.
Meanwhile, for example, Japanese unexamined patent publication No. H4-65942 discloses a technique that does not rely on a centralized control system. In this system, each node has a plurality of predetermined standby routes and normally monitors the state of the standby routes. In the case of a trouble, such a node selects a route appropriate for the state of trouble which has been defined in advance.
Further, for example, in a system disclosed in Japanese unexamined patent publication No. H6-62001, a control frame is generated in the case of a trouble and is transmitted to all routes connected to allow the nodes that receive this control frame to determine a standby route.
In addition, according to the PNNI protocol based on "Private Network-Network Interface: Specification Version 1.0 (PNNI 1.0) (March, 1996)" defined by ATM Forum Technical Committee which is the technical background of the invention, there is defined a routing means for allowing each node to exchange network topology information independently and a signaling means for allowing it to set a connection, and thus lines are connected using a dynamic source routing system. Therefore, in the case of a trouble, the line having the trouble is disconnected.
Among the above-described conventional active/standby routing systems in an ATM network, according to the method of managing network topology on a centralized basis, in the case of an active trouble in a relevant bypass route during operation of the network, it is necessary to calculate another alternative route. Therefore, this method has a problem in that it takes long time to carry out calculations at a large scale and it also takes long time to process information.
The reason is that it is required to calculate bypass routes for all combinations in even a large-scale network in advance, which results in complicated algorithm. Further, centralized management requires various kinds of information in a large amount and requires a long time to collect and process such information.
The above-mentioned method disclosed in Japanese unexamined patent publication No. H4-65942 has a problem in that a large amount of processing is required when a change is made to the configuration of a network during the setting of standby routes which are prepared in advance. The reason is that a change in the network configuration inevitably results in some change in a table which is maintained by every node that forms a part of the network.
The above-mentioned method disclosed in Japanese unexamined patent publication No. H6-62001 has a problem in that it takes long time to complete the switching operation. The reason is that many processing steps are involved because a control frame is generated in response to the occurrence of a trouble to initiate route searching and then the route switching follows.
In a network utilizing the PNNI protocol for connecting lines on a dynamic source routing basis which is the technical background of the invention, it is advantageous to employ a technique to set a standby connection after the starting node sets an active connection and to perform a process of switching the active connection to the standby connection with a small delay in the case of a trouble in the active connection. There is however a problem in that such a switching process with a small delay is impossible. The reason is that although the starting node and end node must know a standby connection for allowing a certain active connection to recover from a trouble in advance in order to perform the switching process with a small delay, conventional signaling does not include information indicating such a pair of active and standby connections.
It is an object of the present invention to provide an active/standby routing system in an ATM network wherein the above-described problems are solved; time required for switching a relevant active route having a trouble to a standby route during active operation of the network is reduced; and the scale of calculations or information processing is reduced to reduce processing time even when another alternative route is to be recalculated because of a trouble in the relevant bypass route.